Gassco supplies Norwegian natural gas to the European market through nearly 8,000 km (5000 miles) of large-diameter high-pressure subsea pipelines. The Norwegian export pipelines are between 300 and 900 km (200-560 miles) long, and have diameters up to 44 inches. Pressure transmitters, flow meters and temperature measurements are only located at the inlet and at the outlet. The state of the gas between those two points can only be predicted by computer models and simulators, which are very important in order to obtain optimal operation of the pipelines. One of the parameters calculated by these models is the temperature profile along the pipeline. This is a very important parameter when assessing the technical integrity of a pipeline by evaluating thermal expansion and buckling. The results from such evaluations are used to decide whether or not rock dumping is needed to stabilize certain parts of the pipeline. The cost associated with rock dumping is typically around 4MNOK (0.5 mill USD) per 100m.
In this study we are comparing three different commercial pipeline simulation tools. The focus is on the ability of predicting gas temperature profiles during steady state operations. The study was initiated after observing quite large deviations between gas temperatures measured by a pipeline inspection tool (PIG) and simulated gas temperature profiles for two different pipelines in operation. In addition, noticeable deviations were observed among the different simulation tools. In order to resolve these issues a series of simulation cases were proposed, all running on a simplified pipeline model.
The results obtained are surprising: after updating and aligning the input model parameters for the real pipeline, all three tools are predicting a much more rapid temperature drop than the measured temperature profiles from the PIG indicate. There are also deviations among the simulation tools, with the first tool (1) predicting a somewhat smaller temperature drop than tool no 2 and 3. On the other hand, the simulation results for the simplified pipeline model shows consistent behavior for simulation tool no 2 and 3. Both in the case of pure adiabatic cooling and high heat transfer from the ambient environment, the results comply with simple analytical models. For these cases, simulation tool no 1 is predicting a smaller temperature drop. The observed differences among the simulation tools might be related to the EoS and/or the chosen strategy for solving the energy equation.
Based upon the results for the simplified pipeline model it is reasonable to question the ability of the PIG to measure a representative gas temperature profile. An initial investigation of the PIG itself and the mounted temperature sensor reveals several possible design flaws that might prevent the sensor from measuring a representative gas temperature. These possible design flaws will be checked in more detail in the future.